HOW MUCH PROFIT
Is Your Current Plan Leaving
ON THE TABLE?An
Enterprise Master Plan Can Show You...and Assure it Never Happens Again
WHITE PAPER WITH CASE STUDY
INSIGHT, Inc. is a world leader in providing supply chain software and consulting services.
Our commercial and custom solutions are designed specifically to meet daunting challenges
in a dynamic, global marketplace: strategic supply chain design, strategic
sourcing/outsourcing, risk management, mergers and acquisitions, optimal production plans
and schedules, transportation procurement and operations, detailed facility operations
analysis, and so on. We relentlessly focus on enabling our clients to outsmart their
competitors and deliver verifiable bottom-line results.
Contact: Jeff Karrenbauer,
President, 703 956 1423 or [email protected]
Arkonas is a management consulting firm specializing in Performance Management.
Through years of unparalleled experience in various industries, Arkonas helps companies
of all industries maximize their profitability by using state of the art methodologies. Contact
John Miller, President; 832 628 7630 or [email protected]
Dybvig Consulting is a boutique firm specializing in the integration of two analytical
approaches: predictive analytics and prescriptive optimization techniques with activity-
dybvig
CONSULTING
based costing data to create an EMP model. The client's EMP model maximizes the ROI
of its total sales and marketing expenditures by simultaneously optimizing the forecast for
maximum profit and the supply chain for optimal feasibility. Contact Alan Dybvig,
Managing Partner; 609 947 2565 or [email protected]
The authors wish to thank Glenn Sabin, Managing Principal of ZS Associates'
office in Princeton, N.J., for his essential contribution to our effort—specifically,
the technical practicality of enterprise response functions. Mr. Sabin can be reached
at [email protected].
Table of Contents
1. Summary
a. Introduction..............................................................................................................4
b. EMP Activity-based Costing Value Proposition ...................................................5
c. EMP Action Plan .....................................................................................................5
2. Current Generation Forecasting and Activity-Based Planning
a. Forecasting ...............................................................................................................7
b. Activity-based Costing ..........................................................................................10
c. Activity-based Planning ........................................................................................10
i. Overview............................................................................................................10
ii. Limitations in practice.....................................................................................11
3. Next Generation Activity-based Forecasting & Planning: Enterprise
Master Plan (EMP)
a. Overview: Optimized Forecasting and Planning ................................................13
b. Building an EMP model .......................................................................................13
i. EMP model “calculation engine” ..................................................................13
ii. Data required for an EMP model...................................................................14
c. Activity-based Plannng Functionality: A Comparison........................................19
4. EMP works
a. Proof of Concept ABC Case Study.......................................................................20
b. Results....................................................................................................................21
5. Conclusion .................................................................................................................22
Appendix
(See John Miller or Alan Dybvig for a draft of Appendix V.)
I. Forecast Process from Future Ready ........................................................................23
II. Activity-based Planning: Details from CAM-I’s The Closed Loop ..........................27
III. EMP model Traditional Cost Functions.................................................................33
IV. Prescriptive Solutions; a numerical illustration of their necessity ........................36
V. EMP model “Proof of Concept” Case Study: Details
a. Company and Financials
b. EMP model Structure
c. EMP model Data
3
1. Summary
a. EMP Value Proposition
This white paper elaborates considerably on an article published in the May/June issue of
Wiley’s Journal of Corporate Accounting and Finance titled “Enterprise Master Plan (EMP):
Next Generation Planning with Activity-Based Costing.”
The elaborations include:
i. A more complete explanation than available in the Wiley article of the EMP value
proposition; specifically that of creating an optimized projected income statement.
This is accomplished by designing, simultaneously, both:
1. An enterprise forecast that is maximally profitable, identifying the profit that the
current forecast is leaving on the table, something never before possible. See
Exhibit 1 below for how an EMP is created.
By
simultaneously
Optimizing Forecast
for
maximum profit
and
Creating
your
To create
Enterprise
your
Optimizing Supply Chain
to guarantee it can make
and fufill the new forecast,
sustainably
Master Plan
Creating an EMP: Exhibit 1
Further, by optimizing the traditionally developed projected income statement, the
Enterprise Master Plan (EMP) assures all the enterprise’s other annual planning
applications are executing to the maximally profitable forecast with the optimally
feasible supply chain. This includes financial (FP&A), operational (S&OP) and
marketing & sales (marketing mix-modeling and sales resource optimization)
applications. In so doing, the EMP also assures all the functional silos are
harnessed to the maximally profitable forecast.
2. The optimally feasible and sustainable supply chain required to make and fulfill
the new forecast.
ii. More details of the EMP “proof of concept” (POC) case study. The authors built a
“proof of concept” EMP model using data from an earlier ABC engagement and
demonstrated the firm had left an additional 25-150% profit opportunity on the table,
depending on the scenario. See Appendix V.
4
b. EMP’s Activity-based Costing Value Proposition
This white paper describes three dynamics that are of unique importance to both activitybased costing software providers and activity-based costing consultants. Specifically:
i. First, the Enterprise Master Plan (EMP) value proposition. It is simple; it assures
all the enterprise’s annual planning applications are executing to the maximally
profitable forecast with the optimally feasible supply chain. This includes financial
(FP&A), operational (S&OP) and marketing & sales (marketing mix-modeling and
sales resource optimization) applications. In so doing, the EMP also assures all the
functional silos are harnessed to the maximally profitable forecast.
ii. Second, as the authors discovered, to their very pleasant collective surprise, the cost
data architecture of an enterprise master plan model (EMP model) and an ABC
model are the same. Thus, EMP models are much more easily built, given a
previous ABC model exists. Specifically,
activity consumption rate (acr) x resource consumption rate (rcr) x the cost factor (cf)
= slope of the requisite EMPmodel cost function curves.
This significantly extends the usefulness of many existing, as well as future,
activity-based cost models, as is described in more detail below.
iii. Finally, EMP models work. They do so by determining, mathematically, how much
profit the firm left on the table because the projected income statement, as
traditionally developed, assumed a fixed forecast and a fixed supply chain. Using
data from an earlier ABC engagement, the authors built a “proof of concept” (POC)
EMP model and demonstrated the firm had left an additional 25-150% profit
opportunity on the table, depending on the scenario.
c. Action Plan
The proposed action plan for interested ABC software providers and consultants is:
i. They become partners with INSIGHT, the software company whose product,
INSIGHT Enterprise Optimizer, creates an EMP model.
ii. The first EMP model to be built will use either last year’s actuals or the most recent
activity-based model data. Further, it will be either:
1. A simplified POC model like the case study described in more detail below, if
the client isn’t completely persuaded of the EMP value proposition, or
2. A full blown EMP model calibration model if the customer decides to bypass the
POC model. This model can use either qualitatively developed enterprise
response functions or ones developed quantitatively by an outside service
provider. As described in more detail below, response functions are the means by
which the assumption of a fixed forecast is relaxed.
iii. The next model after the calibration model will be the calibration model updated
with next year’s planned forecast and supply chain (i.e., projected income
statement), as well as next year’s enterprise response functions quantitatively
developed. This model will then be optimized.
5
iv. The optimized forecast and supply chain developed by the optimization will be
used to update the software containing the original projected statement. From here,
everything proceeds normally. The only difference is now the projected income
statement contains the maximally profitable forecast and the optimally feasible
supply chain required to make and fulfill the new forecast. The EMP model is in
effect, a “back office” activity that the customer’s financial and operations staffs
never see. Thus, it leaves all the customer’s installed financial, operational and
marketing/sales applications in place.
v. The EMP model gets rerun either:
1. Current year when
i. Customer changes the forecast during the year (e.g., with a rolling forecast)
and/or
ii. Variance analyses of the response function(s) determine they need to be
updated
2. The following year when the client’s planning process develops that year’s
projected income statement with a new forecast and updated supply chain data
if required.
6
2. Current Generation Applications
a. Forecasting
Beyond Budgeting Round Table is at the heart of a movement that is searching for ways to
build lean, adaptive and ethical enterprises that can sustain superior competitive
performance.The BBRT is an international shared-learning network of member
organizations with a common interest in transforming their performance management
models to enable sustained, superior performance. For more details, see BBRT.
One of the central performance management tenants of BBRT is that a quality forecast process
is essential. Steve Player, chairman of BBRT NA and co-author of Future Ready: How to
Master the Business Forecast, has very succinctly described just such a process (See Appendix
1). Included are the important distinctions between strategic and execution forecasts ,
business as well as clarifications between goals, budgets and forecasts. ”Business forecasting
takes place when it is possible to steer the business within the constraints of existing goals,
scope and structure of the business.”
A forecast process approach the BBRT has emphasized is that of the Rolling Forecast.
Our concern in this white paper is with business forecasts. Business forecasting is described
by Morlidge and Player, Future Ready: How to Master the Business Forecast, (p. 67). We chose
this name, business forecast, because, while the short term or execution forecast primarily
concerns those that are required to deliver goods and services, and strategy is primarily the
job of senior management, the business horizon usually involves the entire organization in
some fashion.”
Given a high quality forecast process, what are the various techniques by which a forecast
can be created? Referencing Future Ready, (pages 87-124),
There are three types of models can be used to produce a forecast...
1. Despite the disapproval of professional forecasters in academia, the majority of
business forecasting and budgeting processes rely on judgment techniques....
2. The second type of forecast model is the mathematical model...Many businesses
use sophisticated mathematical modeling to forecast volume, perhaps factoring in
the effect of weather on the size of the market or advertising on market share...
3. Given a reasonable amount of historical data, we can use the third type of model: the
statistical (i.e., extrapolation) model. Statistical models employ extrapolation
techniques to generate forecasts.
Another characterization of the differences between mathematical models and extrapolation
models can be found in Hanssens, Parsons, Schultz, Market Response Models, pages 377-378,
386-389. Quoting:
7
“Extrapolative forecasts use only the time series of the dependent variable. Thus, a sales
forecast is made only on the basis of the past history of the sales series...Explanatory (i.e.
mathematical) forecasts go beyond extrapolative by including causal factors thought to
influence the dependent variable of interest.”
In addition to Morlidge & Player and Hanssens et al, explanatory forecasting is also discussed
in Charles Chase’s Demand-Driven Forecasting, second edition, 2011. The process described
below, relaxes the assumption of a fixed forecast by employing what the author characterizes as
“demand-sensing” techniques, more typically referred to as response functions. The solution
is not optimal, however, because descriptive techniques (what will happen if we do “X?”) and
not prescriptive techniques (i.e., what is best “X?”) are used to develop the new forecast.
Demand-driven Forecasting Process
1. Demand Sensing: Uncover market opportunities and key business drivers
(sales and marketing)
2. Demand Shaping: Using what if scenarios, demand planners shape future
demand based on sales/marketing plans
a) optimize sales and marketing tactics and strategies (sales and marketing)
b) assess financial impact (finance)
c) finalize unconstrained demand forecast (sales and marketing)
3. Demand shifting: Match unconstrained demand to supply
a) consensus planning meeting (sales, marketing, finance and operations)
b) rough cut capacity planning review (operations)
4. Demand Response: Constrained demand used to develop supply plan
a) revised demand response (sales and marketing)
b) create supply response (operations)
Forecasting Process from Demand-Driven Forecasting: Exhibit 2
Another explanatory forecast process is described in Hanssens et al, ibid, pages 16-17 and 390396.
Finally, explanatory or mathematical "business" forecasts have also been used for decades
within the sales and marketing functions to size and allocate their respective resources,
optimally.
A comparison of explanatory and extrapolative forecasting techniques is illuminating
(next page).
8
Application
Marketing Mix
Modeling
Sales Resource
Optimization
Business Forecast
Extrapolative
Business Forecast
Explanatory
Planning issue
Size and allocate all
or a portion of
planned marketing
budget
Size and allocate
all or a portion of
planned sales
force budget
Develop a
product(s) forecast
Develop a
product (s)
forecast
How forecast
developed
Multiple time series Multiple time series
One time series
Multiple time series
Marketing plans
drive forecast (i.e.,
they are independent
variables)
Yes
Yes
No
Yes
Marketing response
functions required
Yes
Yes
No
Yes
Forecast’s use
Within marketing
Within sales
Within enterprise
Within enterprise
How forecast
optimized
Prescriptively
Prescriptively
n/a
Descriptively (i.e.,
scenario analysis)
Profit proxy:
Profit proxy:
Objective function contribution margin contribution margin
by product
by product
n/a
Profit proxy:
contribution margin
by product
Best possible
forecast, financially
No
No
n/a
No
Best possible
forecast,
operationally (e.g.,
observe constraints)
No
No
n/a
No
Reference Articles
Hanssens, Parsons, Sinha and Zoltners,
Morlidge and Player,
Schultz, Market “Sales-Force Models:
Future Ready, pages
Response Models,
insights from 25
110-112
“Integrating Market
Years of
Hanssens, Parsons,
response Models in
Implementation,
Schultz, ibid, pages
Sales Forecasting at
Interfaces 31:3,
316-316, 377-378,
Polaroid,” pages
Part 2 of 2,
386-389
391-393
May-June 2001
D.M.Hanssens,
"Order Forecasts,
Retail Sales and the
Marketing Mix for
Consumer
Durables", Journal of
Forecasting, JuneJuly 1998
Statistical Forecasting Applications: Exhibit 3
9
This comparison indicates very clearly the shortcomings of current explanatory and
extrapolative forecasting applications; most significantly, the absence of an enterprise-wide
forecast that is optimal (i.e., maximally profitable). As described below, that shortcomings
are eliminated with an Enterprise Master Plan. Next generation forecasting and planning is
here. Now.
Right now.
b. Activity-based Costing
“Activity-based costing was first clearly defined in 1987 by Robert Kaplan and W. Bruns in a
chapter in their book Accounting and Management: A Field Study Perspective… During this
time, the Consortium for Advanced Management-International, now known simply as CAM-I,
provided a formative role for studying and formalizing the principles that have become more
formally known as Activity-Based Costing.” Wikipedia.
The adoption of ABC techniques was not without its challenges, however.
“Companies rejected ABC on the basis of its perceived administrative and technical
complexity and its need for new systems continuously generating activity data. While ABC
model is feasible for initial pilot studies, it is difficult to extend to company-wide
applications. Even after the initial model has been built, updating the model requires
essentially re-estimating through a new round of interviews and surveys to reflect changes
in company’s operations. Consequently, ABC models are often not maintained and their cost
estimates soon become obsolete (Kaplan 2003).” Velmurugan, Journal of Performance
Management, May 1, 2010.
However, a subsequent development, time-based activity based costing, addressed many of
these issues and has been successfully commercialized by Acorn Systems Pilbara. Today,
there a variety of commercial ABC offerings available from both larger performance
management vendors (e.g., Cognos, Infor, Oracle, SAP, SAS) as well as standalone firms
(e.g., Acorn Systems, Decimal, Prodacapo).
c. Activity-based Planning
i. Overview
It didn’t take long, after the initial promulgation of ABC concepts, for academics and
practitioners to turn their attention from the use of ABC techniques to define customer
and product profitability more accurately using last year’s results (metaphorically, the
back of the planning boat) to turn their attention to applying the same concepts and data
to planning the next year’s results (i.e., the bow of the planning boat). The essential ABC
planning factors, activity consumption rates, resource consumption rates, and cost factors
remained the same. The only difference was that data in the model flowed in the
opposite direction: from products and customers through activities to resources rather
than from resources through activities to products and customers.
10
A variety of books were published describing this “reverse flow” (sometimes, also,
referred to as activity-based budgeting) Examples include Kaplan and Anderson, TimeDriven Activity-Based Costing (Chapter 5), 2007; Kaplan and Cooper, Cost & Effect (Chapter
15), 1998; Cokins, Activity-Based Cost Management (Chapter 8), 2001; and, finally, Hansen
and Torok, editors, The Closed Loop, Implementing Activity-based Planning and Budgeting.
CAM-I, 2004.
Of all these books, CAM-I’s The Closed Loop is the most detailed. (See Appendix II for a
summary of The Closed Loop’s process). Thus, its formulation of activity-based planning
will be used for the remainder of this white paper to represent “best practices” for the
current generation of activity-based planning efforts.
The activity-based planning model the book describes, the Closed Loop model (CL
model), will be used in comparison with the next generation activity-based planning
model described in the remainder of the white paper, the Enterprise Master Plan model
(EMP model).
Summarizing, editors, authors and contributors of The Closed Loop held that:
“The Closed-Loop and the Activity-Based Budgeting and Planning Process are the
most significant development in the field of Planning and Budgeting in the last
thirty years.”
Also, “We have developed a planning and budgeting approach that extends activitybased logic into a new domain: planning and budgeting.”
And, “In the long run, a successful organization will switch from a primary focus on
generating budgets to a more fruitful focus on planning.”
ii. Limitations in the Current Practice
However, as recognized by the editors, authors, and contributors, “While the
concepts are straightforward, performing the necessary calculations is fairly intricate.”
Examples include:
a. Too many options (“moving parts”) within several of the 7 steps (See Appendix II).
i. 12 options in Step 3: Balance Resource Requirements with Resource Supply
ii. 13 options in Step 6: Balance Financial T Results with Financial Targets
b. Many options repeat themselves across the steps
c. Too many sequential steps in the overall process: 7
d. No simultaneity since “calculation engine” of the CL model cannot consider any of
the steps or options at the same time
11
e. No standard software existed to build the CL model calculation engine and store
the associated data.
f. “The practical difficulty in comparing resources supplied and resources required
occurs when the unit of measure used for supply of a resource differs from the unit
of measure in which the resource is used. (e.g., requirements for people are often
expressed in hours whereas people are generally acquired in FTEs.)”
As will now be demonstrated, all these shortcomings have been eliminated with the
Enterprise Master Plan. Next-generation activity-based forecasting and planning is here.
12
3. Next Generation Forecasting and
Activity-based Planning: Enterprise Master Plan
a. Overview: Optimized Forecasting and Planning
There are five factors necessary for developing a maximally profitable annual plan
(i.e., projected income statement):
1. Forecast must be a variable in the activity-based plan model
2. Supply chain must be variable in the activity-based plan model
3. Objective function (i.e., what you're trying to optimize) must be profit
4. Solver must be prescriptive (“what is the best X?”) and not scenario analysis
(what will happen if we do “X”?) (See Appendix IV for an illustration)
5. Solution must be developed with a simultaneous consideration of all the variables.
One or more of these factors are used in most planning software available today. The
EMP model is the only one, however, which incorporates all five factors. The EMP
model accomplishes this using new planning software that integrates three planning
techniques which have been commercially available for decades. The three are: i) supply
chain network design, ii) activity-based costing and iii) marketing-mix modeling.
• The supply chain software relaxes the assumption of a fixed supply chain (details
below), uses profit as the objective function and has a simultaneous prescriptive solver.
• The activity-based costing software provides the data for the cost functions in an EMP
model by which the assumption of a fixed supply chain are relaxed.
• Finally, the assumption of a fixed forecast is relaxed by the EMP model employing
enterprise response functions, developed in traditional marketing-mix modeling
software.
b. Building an EMP model requires the following:
i. EMP model “calculation engine”
The software used to create an EMP model is that described by Jeff Karrenbauer,
President of INSIGHT, at the Fall, 2013 CAM-I meeting held Naperville, IL on
September 10, 2013. A copy of the presentation is available upon request. The
Agenda was
i) Supply Chain Management- Myth vs. Reality,
ii) Supply Chain Management-An Analytic Perspective,
iii) Strategic Sourcing,
iv) Sales and Operations Planning (S&OP),
v) Unification of Marketing and SCM
vi) SCM and the Green Movement
13
ii. Data required for an EMP model
a. Structure
As is true of any supply chain network design (upon which an EMP model is based), the
model structure of the projected income statement is a series of geographically-located
nodes connected by links arranged in a hierarchy, procurement to customer. The nodes
contain facilities and within the facilities, activities and products. These nodes and links
are appropriately constrained (e.g., capacities).
However, the flows within the network (e.g., across a node, within a facility, through an
activity) are not known because they are the answer to the question: “What is the
optimal supply chain configuration to make, fulfill and service the forecast?” Thus, the
essential requirement for optimized planning is an understanding of unit costs and how
they vary with volume. As will be described below, these relationships are referred to as
cost functions.
In addition to structure, the key data elements of an EMP model are: cost functions,
capacity and related constraints, and demand, all of which can be are obtained from an
ABC model or CL model. The final elements, enterprise response functions, are
provided by outside experts.
b. Cost functions
As described above, all the network costs in any EMP model must be represented as cost
functions.
Cost functions are defined by Dr. Charles Horngren as “descriptions of how a cost changes
with changes in the level of an activity or volume relating to that cost.” Cost functions describe,
mathematically, the relationship between activity changes (units, weight or volume) and
the cost changes driven by the activity changes.
Cost functions must be a combination of fixed and/or linearly variable volumes, given
the mathematical programming techniques that are used to optimize the EMP model.
These include:
• linearly variable with increases or decreases in activity
• Fixed costs that don’t change with activity at all.
• Stepwise fixed
• Any combination of fixed and linear
14
Thus, plotting the cost function with changes in cost on y axis (dependent variable) and
changes in units of volume on x axis (independent variable) yields the following:
cost = slope x activity. The slope is expressed as cost/activity and is the key mathematical
factor in the cost functions.
Traditionally, there have been 3 different approaches used by the supply chain community
to develop cost functions: i) accounting, ii) statistical and iii) engineering. (For more details,
see Appendix III)
However, fortunately, (though not well understood), two analytic techniques of interest
(activity-based costing and supply chain network design) have exactly the same costing
data architecture. This, in a nutshell, is why EMP models can be easily created from ABC
data.
Reviewing, activity consumption rate (acr = activity/product) and resource consumption
rate (rcr = resource/activity) and the associated cost factor (cf = $/resource) when multiplied
are, in fact, precisely the slope of the variable cost functions required in an EMP model.
Thus:
slope = activity/unit of product x resource/activity x $/resource = $/unit of product = slope of
cost function curve.
Below is a graphic describing the use of the three ABC factors in The Closed Loop
planning process flow. Reiterating, activity-based planning flows products/customers
through activities to resources; activity based-costing processes flow in the opposite direction.
Exhibit 4
15
Obviously, as illustrated above, the three essential factors of The Closed Loop are exactly
those developed by traditional activity-based costing techniques: the activity
consumption rate (acr), the resource consumption rate (rcr) and cost assignments (ca).
They are also sometimes referred to as cost factors (cf).
In The Closed Loop, this is illustrated in Chapter 8 with an example; that of an outbound
call center. All the costs in the example are fixed except those of the reps making the
calls and the telecom costs/call.
Using data from the call center example in The Closed Loop, below is an illustration of the
arithmetic identity of the slope of the cost function curve required in an EMP model and
the multiplication of the three factors developed in the activity-based costing analysis:
acr x rcr x cf= EMP model cost function slope.
Mapping Call Center Consumption Rates to Cost Functions:
LABOR
Cost Object = Campaign
Activity = Create Campaign
Activity Demand
(calls/campaign)
1. Activity Consumption Rate (ACR)
ACR = 100K calls/campaign
200K
100K
1
2
Activity (# Campaigns)
2. Resource Consumption Rate (RCR)
Resource Demand
(FTE/campaign)
RCR = 10 min./call
Assume: FTE = 1500 hrs.
FTE/Campaign = 10 calls/campaign
x 10 min./call
x 1 hr./60 min.
x 1FTE/1500 hours
= 11.1 FTE/campaign
133.3
22.2
11.1
1
12
2
Activity (# Campaigns)
3. Assume: FTE = $50K
Campaign cost = $50K/FTE
COST FUNCTION (LABOR)
Resource Supply Capacity
$7.5m
$6.7m
Resource Demand
(cost)
x 11.1 FTE/campaign
= $555K/campaign
Activity
Capacity
$1.11m
$555K
1
2
12 13.5
Activity (# Campaigns)
Developing an EMP model Cost function from Closed Loop ABC Data: Exhibit 5
16
c. Capacity and Other Constraints
All constraints, including capacities, must be identified as they are an explicit
requirement for optimization. Further, in most cases, these constraints can be relaxed.
Examples include:
• Limits on procurement availability
• Manufacturing capacity
• Sales and marketing expenditure limits
• DC throughput, storage
• Energy consumption
• Carbon emissions
• Targets for inventory and customer service
• Transportation link restrictions
• Supply/demand imbalances (e.g., inventory build ahead vs. over time)
d. Enterprise Response Functions
Response functions have been around for decades and link sales or marketing activities
to forecast/revenue results. Specifically, they relax the assumption of a fixed forecast by
predicting volumes/revenues at different levels of sales or marketing effort. Sales
response functions are used to size and allocate the sales force resource (sales resource
optimization (SRO))while marketing response functions are used to size and allocate the
marketing budget (marketing- mix modeling (MMM))
Response functions are the reverse of cost functions because the independent variable is
not units but rather sales and marketing expenditures. The dependent variable is units.
Units are, also, frequently multiplied by price to yield revenues as the dependent
variable.
These relationships have been traditionally used to inform critical resource allocation
decisions including how big the sales or marketing budget should be, and to which
products and/or customers should these resources be allocated. As a result, this process
can lead to changes in individual product or customer expenditures.
In these approaches, the supply chain is fixed and the objective is to maximize the
contribution of the sales and marketing efforts after accounting for the costs of these
promotions and a fixed product margin. It is not common to account for changes in
margin as a function of the expected product demand.
17
There are a broad range of methods that can be used to estimate response functions,
which differ in the time/ effort involved and the precision that can be achieved. A partial
list of these methods includes:
• In-market tests to isolate the impact of individual promotions
• Econometric methods that rely on statistical analysis to estimate the sales impact of
prior sales and marketing activities
• Expert sessions that provide a structured process to solicit and refine estimates of the
impact that a promotion will have
Regardless of how the response functions are derived, they can be compared to actual
results and re-calibrated as needed. This is analogous to the financial variance analysis
process.
In conclusion, it is the extension of sales and marketing response functions to enterprise
response functions that enables the development of an optimal forecast that is maximally
profitable, something never before possible.
18
c. Activity-based Plannng Functionality: A Comparison
Comparison Factors
CL model
EMP model
1. Steps in Activity-Based
Planning Process
Step #1
Develop demand
statement
Same
Steps #2, #3, #4, #5, and #6
Performed sequentially
Performed simultaneously
Step #7
Create plan
Same
2. Data
Traditional ABC planning
factors of acr, rer, cost
factors, demand and
capacities
Same (80+%) plus response
functions
3. Solver
Scenario analysis (what will
happen if we do X?)
Prescriptive (What is the best
possible X?)
4. Results
Sub-optimal forecast,
supply chain and profit
Maximally profitable forecast
and optimallyfeasible supply
chain, assured
5. Modeling software
“calculation engine” didn’t
exist when Closed Loop
published
INSIGHT’s INSIGHT
Enterprise Optimizer (IEO)
6. Additional software
functionality
None
Sustainability (energy and
carbon emissions)
Comparison: EMP model and CL model Functionality: Exhibit 6
19
4. EMP model Works
a. Proof of Concept case study (See Appendix V for details)
It started out as a simple comment six years ago, “Imagine relaxing the assumption of a fixed
forecast to solve for the optimum level of sales and marketing investment that provides the
highest profit and ROI.” Planning and Budgeting, Arkonas One Eighty Newsletter,
February, 2008
The software required to accomplish this, INSIGHT Integrated Enterprise Optimizer (IEO),
was already under development at INSIGHT, a provider of software used for optimizing a
supply chain network. The concept was simple: Using IEO, create an EMP model of the
current projected income statement as traditionally developed and, then, optimize the ROI of
its total sales and marketing expenditures. The resulting Enterprise Master Plan (EMP)
produces the maximally profitable forecast that the projected income statement’s resources
are capable of making and fulfilling. Simultaneously, enterprise-wide, IEO resizes and
reallocates these same resources to support the manufacture, fulfillment and support of the
new forecast; i.e., the supply chain is assured to be optimally feasible.
But would it work? Would it actually demonstrate a substantial profit improvement? It was a
difficult question since no firm had been found willing to proceed without credible proof that
it would deliver what it promised. In other words, a “proof of concept” (POC) model was
required. Rather than inventing data, it made more sense to find an existing set of actual data
and use it to create the POC model.
The modeling results most readily available were of a previous ABC engagement conducted
by one of the authors. Fortunately, the data incorporated the entire income statement. So, an
investigation was made into the match between the ABC data developed and the data
requirements for an EMP model. Specifically, whether the EMP model POC cost function
slopes could be developed from the ABC data.
Using data from the call center example from CAM-I’s The Closed Loop, what was learned was
very important. (See page 13) The two seemingly unrelated, activity-based analytic
techniques — activity-based costing and an EMP model — share common activity-based
costing data architecture. Both techniques build their model with fixed and linearly variable
relationships between costs and activity (units, weight or volume).
20
b. Results
Table 1: McCoy Company Results (Far East 20%)
Scenarios
Maximize
Revenue
Maximize
Profit
Baseline
$136.3 m
$12.7m
Revenue
max
$143.8m
(6%)
Profit max
$140.9m
(3%)
Sales/Marrketing
Sales/Marketing
ROI
Activity
capacity
exceeded
$28m
45%
None
$16.3m
(28%)
$28.6m
27%
improvement
1 (labor)
$19.8m
(56%)
$23.6m
87%
improvement
1 (labor)
Table 2: McCoy Company Results (Far East 200%)
Scenarios
Maximize
Revenue
Maximize
Profit
Baseline
$136.3 m
$12.7m
Revenue
max
$173.4m
(6%)27
Profit max
$170.5m
(25%)
Sales/Marrketing
Sales/Marketing
ROI
$28m
45%
$30.0m
(136%)
$34m
96%
improvement
$33.5m
(164%)
$39m
158%
improvement
Activity
capacity
exceeded
None
5 (labor)
2 machine
4 (labor)
2 machine
21
5. Conclusion
As described above, an EMP model solves the two biggest impediments limiting the
commercial success of activity-based planning.
• It eliminates the “intricacies” of the Closed Loop process and CL model’s “calculation
engine”
• It assures the forecast is maximally profitable and the supply chain optimally feasible.
It accomplishes this by simultaneously and optimally balancing supply, demand, profitability
and the supply chain.
Thus, EMP model’s functionality truly represents the next generation ABC-based planning,
both financially and operationally. Further, it does not employ “new” or “untested” analytics.
Rather, it is simply the integration of three different and robust sets of analytics (i.e., mixed
integer and linear math programming, predictive analytics and activity-based costing) that have
been commercially successful for decades.
Also, for firms and consultants whose experience is with ABC modeling, the EMP model is a
platform that extends the operational uses of ABC data from efforts focused on process
improvements and customer/product profitability (i.e., the back of the boat) to planning
applications like forecasting, finance, operations and sales/marketing (i.e., the bow of the
boat).
Finally, an ABC EMP model can be built with relatively little additional data gathering, as
described above. This introduces the client to next generation activity-based forecasting and
planning while simplifying the EMP model build effort significantly.
22
increasing choice
Operational
Forecasting
Business
Forecasting
Strategic
Planning
ADAPTATION
Freedom of action
Alternative scenarios of the future
environment
Broad brush estimates
ADAPTATION
How do we structure the
business to compete
most effectively?
Choice of response limited
Best estimate of what will happen
(based on current assumptions)
Detailed enough (with ranges)
How do we deploy our
resources to best effect?
Highly constrained
Prediction of what will happen
Detailed forecasts
How do we service
demand efficiently?
Implementation
RESPONSE
RESPONSE
Decision Making
NAVIGATION
NAVIGATION
Creating Options
PROCESS
PURPOSE
Appendix I: Forecast Process from Future Ready
increasing predictability
23
24
T = timely
A = actionable
R = reliable
A = aligned
C = cost Effective
Specification for a forecast
TARAC
CARAT
which we change to achieve our target
A set of plans : what we intend to do,
based on:
A forecast - what we think will happen
which we achieve by producing
A target is what we would like to happen
Key concepts
25
..while it is
necessary to have
a plan….
QUARTER
the way you expected so….
…you first need to
work out where
you are heading
(FORECAST)….
…the world often turns out different to
©Satori Partners
Ltd
QUARTER
do differently to get back on
track (TARGET)
..and then what you have to
Forecasting is about decision making, because…
26
We haven’t forecasted in a while,
maybe we should try that again….
Don’t treat forecasting as a “special event” instead of an
on-going part of monitoring the business
Appendix II: Details of activity-based planning process
What follows for the rest of the Overview are direct quotes from The Closed Loop
The book contains the editors, authors and contributors “recommended approach:”
•The CAM-I ABPB Closed-Loop, (referred to as “the Closed-Loop ), a new
approach to calculating the activity, resource and financial requirements of an
organization and it units.
•The CAM-I ABPB Closed-Loop Process, (which contains the business processes
and techniques needed to support the Closed-Loop and
•The CAM-I ABPB Implementation Program, a structured approach to introducing
the Closed-Loop and ABPB Process into an organization
Benefits of the ABPB “recommend approach” included:
1. Reduction in the time and cost of generating a budget (and a plan, authors’ addition)
2. More accurate costs and better decision making
3. More specific and cohesive link with the strategic plan
4. Added ability to adjust activity and resource consumption rates
5. Additional methods to adjust capacity
6. Reduced time to collect information
7. Disagreements become more transparent
8. Decrease in political gaming
9. Easier to communicate and increased buy-in
10. Improved understanding by managers
11. Improved justification for budget requests
12. Superior response to last minute changes in assumptions
13. Improved cash flow forecasts
14. Improved ‘what if” analysis
15. Easier integration with other processes
The heart of the recommended approach is a budget (or planning, authors’ addition)
calculation engine; the ABPB Closed-Loop …The Closed-Loop has three important
features:
• It is activity-based
• It explicitly matches resource demand and resource capacity and
• It achieves operational balance and then confirms financial balance
The Closed-Loop has two stages. The first, the operational loop, balances the demands on
organizational resources and their supply in purely quantitative (but, non- financial) terms…
Balancing resource demand and supply is the most difficult step in any plan and is valuable
even without the inclusion of costs in the planning process.
27
The second stage, the financial loop, adds costs of resources and the value (i.e., revenue) of
output to generate feasible financial plans.
The detailed calculations of stage 1 and 2 involved in the Closed- Loop (CLM) are shown in
Figure 1, above.
Steps in the Closed Loop Process utilizing the Closed-Loop include:
1. Set quantitative demands. Using the organization’s strategy, the quantity of demand in the
upcoming period is estimated for each product or service. This estimation is done purely in
quantitative terms, such as number of units, tons, accounts, customers, shipments, and so on.
2. Determine resource requirements (CLM). Step two in the CLM consists of two distinct
actions:
a. First, the quantity of demand is converted into activity requirements, expressed in
operational terms, using activity consumption rates. Activity consumption rates are
defined as: the number of occurrences of an activity required to generate a single unit of
output.
b. Second, the activity requirements are converted into the individual resource
requirements using resource consumption rates. Resource consumption rates are
defined as: the quantity of each resource required to undertake a single occurrence of
an activity.
3. Balance resource requirements with resource supply (CLM). The most critical aspect of
the CLM is establishing operational balance by matching resources requirements with
resource capacity in a given time period.
28
The practical difficulty in comparing resources supplied and resources required occurs
when the unit of measure used for supply of a resource differs from the unit of measure in
which the resource is used. (e.g., requirements for people are often expressed in hours
whereas people are generally acquired in FTEs).
Once a common capacity measure is determined, management compares each resource’s supply
with its corresponding requirement. This comparison results in one of three situations:
a. Too much capacity: Too much capacity can be operationally feasible, but is only
operationally balanced if the amount of excess capacity is needed as a buffer. If this
is the case, the situation needs to be analyzed to see if it meets financial balance.
b. Too little capacity: Demand requirements cannot be met because there is a
shortage of resources. This situation is not operationally balanced and therefore
cannot be operationally balanced.
c. Exact balance of capacity and demand: This situation is operationally feasible and,
by definition, in operational balance. The situation needs to be analyzed to see if it
meets financial balance.
Armed with the knowledge of any imbalances, management must assess capacity by
determining whether or not:
d. A surplus or shortage of resources is large enough to justify action or if it should
just be accepted.
e. The surplus or shortage is expected to last for a long time. Shortages that continue
for a long period of time can cause excessive costs for overtime or decreased
performance or service levels.
f. The capacity cannot be changed in the time period being addressed,given
prevailing economic conditions.
With a capacity assessment in hand, there are three distinct ways that the organization can
achieve operational balance (see Fig. 1, operational balance on left side of graphic):
g. A surplus or shortage of resources is large enough to justify action or it should be
accepted. NOTE: The CLM is a calculation algorithm. As such, it will treat a
situation where resources required exceed resources supplied as infeasible and
therefore the organization could theoretically not meet its demand requirements.
In extreme cases, this may be true. However, in many cases, the use of overtime,
tEMP model labor, or extremely high machine utilization-especially for short time
29
periods-may be EMP model loyed to meet demand. This case is, in effect, a new
plan with greater capacity and higher costs.
h. The surplus or shortage is expected to last for a long time. Shortages that
continue for a long period can cause excessive cost for overtime or decreased
performance or service levels.
i. The capacity cannot be changed in the time horizon being addressed, given
prevailing economic conditions.
With a capacity assessment, there are three distinct ways that the organization can achieve
operational balance:
j. Adjust capacity or improve usage of resources. With operational balance, the demand
requirements can be met and there is either no excess capacity or an acceptable
quantity of idle or buffer capacity.
k. Adjust the activity and/or resource consumption rates to resolve or at least reduce the
magnitude of the problem. Management usually seeks to implement any available
economic effectiveness or efficiency opportunity.
l. Change the absolute mix of products/services demanded. A well planned product change
may either absorb the excess or reduce the shortage or resources without affecting
the organization’s overall strategy.
4. Determine resource costs and derive financial results (CLM).
To determine resources costs, two elements are required:
a. the unit cost of each resource required and
b. the quantity of each resource as determined from the operational plan.
The financial elements need to complete this step are:
c. Unit cost of each resources, such as the hourly wage or annual salary and
d. Revenue (price) per unit of demand
The cost of each resource is assigned to the resource and using the relationship between
resources and each activity these costs are assigned to the activities requiring or consuming
tat resource. These activity costs are then assigned to products to derive product and
service costs.
30
5. Add non-activity-based costs to obtain the total financial result
Certain organizational costs may not have a direct or tangible correlation with activity
volume and therefore are better handled through a more traditional budgeting approach
than the CLM . These types of costs are sometimes referred to as “business sustaining”
costs. Examples include directors’ fees, certain building leases, SEC filing fees, etc.
Ultimately, all costs (activity-based and non-activity-based) that apply to the organization
must be considered to generate a financial plan.
6. Balance Financial Results with Financial Targets
When a financial plan has been prepared, an assessment is made of whether or not the
total projected financial results meet the required targets of the organization. If these
targets are not achieved, three options can be pursued either individually or in combination
with one another. These options are:
a. Adjust demand pricing, assuming that the new pricing is compatible with the
market. An organization must estimate the impact of an increase or decrease in
price on the quantity of demand.
b. Modify resource costs including the possibility of outsourcing. The second option
is to adjust or eliminate shift premiums or overtime. Options include:
i.Re-calibrating shifts to reduce or eliminate current shift premiums or overtime
ii. Adopting a two-tier wage structure where new EMP model loyees are paid
less than existing ones
iii. Adjusting compensation plans to allow for incentive pay to be more closely
tied to organization results
iv. Increasing or decreasing wages to encourage EMP model loyees to join or
leave the organization
v. Paying for skill rather than seniority
vi. Outsourcing to obtain lower costs from more efficient suppliers or reducing
ecess capacity costs by paying only for what is used
vii. Negotiating more favorable energy and other supply contracts
viii. Substituting less expensive materials, providing that other costs (e.g.,
processing effort or waste) are not increased. As with the approach of changing
prices for products and services, chaning resource unit costs might also affect
resource quantity.. If that is the case, the operational plan must be reviewed to
ensure the new level of resource quantity continues to provide operational
balance.
c. Change one or more of the operational parameters directly:
i.Quantity of demand including mix
ii. Activity and/or resource consumption rates
iii. Available capacity
31
At some point, a satisfactory operational and financial balance will be achieved that meets
the strategic requirements for the organization. At this point, a formal budget (or plan,
authors’ addition) can be generated in the appropriate format and structure for the
organization.
7. Create a Formal Plan
When both operational and financial balance have been achieved, a more formal line-item
budget can be created.
In some organizations it may not be necessary to go to this level of detail. The operational
and financial plans generated by the CLM may be sufficient to run the business.
Alternatively, the formal budget detail may be required but on a less frequent basis. For
example, the CLM could be used on a quarterly basis and the detailed formal budget
generated annually.
We view the APBP Process as having more to do with planning than budgeting.
In the long run, a successful organization will switch from a primary focus on generating
budgets to a more fruitful focus on planning.”
32
Appendix III: Traditional Cost Function Curve Development
1. Accounting Approach
The most popular approach to facility data preparation is based on a detailed analysis of
historical cost accounting records. The basic idea is to assemble all relevant cost accounting
records, remove extraneous information, ensure comparability, separate fixed and variable
costs, perform consistency checks, and prepare final model inputs. If you choose the
accounting approach, we recommend that you follow the step-by-step procedure outlined
below.
Step Action
1 Identify all accounts that contain facility operating costs.
2 Obtain historical data for each account identified in Step 1 for each facility
active during the base period of the study.
3 Identify and remove from each account any costs that are not related to
facility operations.
4 Carefully study reporting standards and practices by facility location. Attempt
to identify discrepancies that would yield misleading results. The basic idea is
to ensure later apples-to-apples comparisons across facilities.
5 Identify and temporarily remove cost differences between facilities which are
due to regional influences (for example, labor and utility rate differentials).
You may wish to use the Regional Cost Indices included in SAILS to facilitate
this effort. This is done to ease Steps 6-8.
6 Separate facilities by generic type and mission. For example, distribution
centers should, at a minimum, be segregated into owned, leased, and public
categories. Use additional subdivisions as required to account for important
operating differences: dry vs. refrigerated, bulk vs. bin, etc. Review the
discussion of noncomparable facilities, as necessary.
7 Analyze carefully the results from Step 6 for consistency across facilities.
Within a given facility type, perform ratio tests such as those described
earlier. If discrepancies are present, you must attempt to explain them.
Remember that you have already accounted for extraneous costs (Step 3),
reporting practice inconsistencies (Step 4), regional influences (Step 5), and
mission differences (Step 6). If discrepancies persist, then you are likely
faced with the delicate (and potentially explosive) matter of managerial
and/or labor force performance deficiencies. Unless you have compelling
evidence to suggest that such problems are inherent, we recommend that
you do not represent them in your model. Choose a representative set of
costs and ignore substandard operating practices. From a strategic point of
view, such variances should not be the basis for a network redesign.
8 Categorize each account as either fixed or variable. (Refer to our earlier
definitions of fixed and variable costs, if necessary.) This step will almost
33
certainly involve some judgment calls on your part.
9 Reintroduce regional differences removed at Step 5.
10 Prepare final inputs for your model.
2. Statistical Analysis Approach
One of the most difficult challenges that you must face when analyzing historical facility
costs is the segregation of accounts into fixed and variable categories. The statistical
approach circumvents this problem because it is completely independent of the nature of
individual cost accounts. The basic idea is to derive a mathematical function that best
describes the observed relationship between total cost and facility volume. The statistical
technique that you will normally use is single variable linear regression.
The statistical approach to facility data preparation is summarized next:
Step Action
1-7 Follow steps 1-7 from Accounting Approach, earlier.
8 Perform regression analysis of total facility costs (dependent variable) and volume
(independent variable). Interpret resulting equation coefficients as follows:
y-intercept: fixed cost
slope: variable costs
9-10 Follow steps 9-10 from Accounting Approach.
These cost coefficients probably will bear little resemblance to those you derive
via the accounting approach. Nevertheless, if the equation fits observed historical
data reasonably well, it is equally valid. Furthermore, you are relieved of
the difficult task of attempting to classify accounts as fixed or variable.
Rather, you are simply asserting that the total cost function for a given facility
type behaves in a predictable, justifiable way; the underlying components of
total cost are unimportant to the solver.
3. Engineering Approach
Suitable historical facility operating costs may not be obtainable from your accounting
records. Even if they are available, you may be unwilling to use them as the basis for your
analysis for several reasons, including
a. base period that contains abnormal events such as strikes or national disasters;
b. reporting discrepancies that are so severe they cannot be reconciled, and
c. missing information from one or more facilities
In such instances, you may conclude that standard costs should be used instead of
accounting data. If your firm has recently built, or plans soon to build a new manufacturing
or distribution center facility, it is virtually certain that the planning phase involved detailed
estimates of facility operating costs. Assuming that the facility size specified in the analysis
34
represents those you wish to evaluate, you can incorporate these values in your IEO model.
Alternatively, you can commission special studies to develop such estimates. Following is
the recommended step-by-step procedure:
Step Action
1 Obtain engineering cost estimates for each facility type to be evaluated.
2 Identify and temporarily remove regional influences built into the estimates
(for example, labor and utility rate differentials). You may wish to use the
Regional Cost Indices to facilitate this effort. You should perform this step
even though you will almost immediately reintroduce such factors in Step 4.
Remember that most engineering studies are confined to few sites. If you wish
to use this data to evaluate a larger number of candidates, then the base cost
estimates must be region neutral.
3 Ensure that your standard costs are divided into fixed and variable components.
Obtain the assistance of the engineering design group responsible for the
estimates, if required, to perform the required segregation.
35
Appendix IV: Prescriptive Solutions; A numerical illustration
of their necessity
An EPM POC case study model relaxes two fundamental constraints in traditional planning
models: that of a fixed supply chain and that of a fixed forecast. Adding response functions to a
POC model to relax the assumption of a fixed forecast increases the number of scenarios that
would have to be run if the solution was to be determined by scenario analysis (i.e.,
descriptively) (NOTE: These scenarios are in addition those required to evaluate the supply
chain scenarios.)
That is the only way to determine, descriptively, which of the various scenarios (which answer
the question: “What would happen if we do X?”) answers the much more important question:
“What is the best X?”
There are three factors in the model which determine the answer to the question. They are
Products (P), Objective functions (OF) and Customers (C).
The number of products and objective functions increases the number of scenarios
multiplicatively. Unfortunately, the number of customers increases the number of scenarios
exponentially. This is because every customer has 2 possible “states:” that of having more
demand purchased for them by the model or not having had demand purchased. Thus, the
total number of customer demand configurations is 2 to the number of customers.
Two examples demonstrate, overwhelmingly, when descriptive solutions MUST yield to
normative for anything like a realistic, actionable model.
1. This yields for the McCoy POC model where P=2, OF = 2 and C = 9, 2 x 2 x 2 to
the 9th (= 512) or 2048 scenarios.
2. For a more realistic model where P = 10, OF = 2 and C = 50, the answer is 10 x 2 x 2
to the 50th (= 10 to the 15th) or 2 x 10 to the 16th.
3. When the numbers of possible solutions are expanded to include relaxing the
assumption of a fixed supply chain, the case for scenario analysis becomes just that
much more absurd. In this case, the integer variables are not absence of presence of
a customer responding to a response function. Rather it is the absence of presence
of a facility, product, activity, etc in the solution. This increases the possible
solutions to, literally, more stars than there are in the universe which is 10 to the
24th.
36